Professional Documents
Culture Documents
Wall Buildings
3.1 Introduction
One of the missions of the Pacific Earthquake Engineering Research (PEER) Center is to
develop procedures and guidelines for performance-based earthquake engineering. The damage
resulting from the Izmit earthquake gave the reconnaissance team a unique opportunity to study
response limit states for selected buildings impacted by the earthquake. This chapter summarizes the
key observations of the reconnaissance team regarding residential and commercial reinforced concrete
construction in Turkey. Construction practices are described and the responses of moment-resisting
frames and the behavior of shear walls are summarized. In the epicentral region, the two most widely
used framing systems for residential and commercial construction are reinforced concrete moment-
resisting frames and shear walls, and most of the loss of life and damage in the Izmit earthquake was a
result of the poor performance of reinforced concrete buildings.
The reconnaissance team found evidence of both extremely good and extremely poor
commercial construction. One example of excellent quality construction is the building of Figure 3-1
that was located in central Yalova. Residential construction quality also ranged from excellent to poor.
Although construction work by registered contractors was generally of better quality than that by
individuals and homeowners, the quality of contractor-completed construction was often poor by U.S.
standards. Figure 3-2 shows a completed shear wall in a multistory
apartment building in Yalova. The vertical and horizontal rebar in
the shear wall can be seen on the exterior face of the wall. Although
buildings in Yalova suffered damage in the earthquake, the
apartment building was not damaged despite the poor quality of the
concrete evident in the photograph.
Figure 3-2 Poor quality construction of a shear wall in Figure 3-3 Homeowner apartment building construction
an apartment building in Yalova
Smooth rebar is commonly used for reinforced concrete construction in the epicentral region.
The yield strength of such rebar is approximately 275 MPa. Smooth rebar is used because it is less
expensive and more readily available than deformed rebar, and is also easier to bend and cut on site.
The strength and quality of the concrete varied widely as noted above. Concrete is typically batched on
site for low-rise residential and commercial construction, and standard quality control procedures such
as slump tests are rarely used. Low-strength concrete was identified in a number of damaged buildings
visited by the reconnaissance team. Some samples were weak enough to crush by hand.
The reconnaissance team was surprised by the volume and type of residential construction.
Specifically, there was much unoccupied new residential construction, and there were many
incomplete single- and two-story additions to existing construction. Local experts explained that
homeowners often added stories to existing apartments or constructed new multistory apartments as a
hedge against inflation. The quality of such construction was often poor; it is highly likely that much of
this construction was neither engineered nor approved by the local jurisdiction.
Figure 3-4 Typical gravity framing including beam and column details
Figure 3-6 presents photographs of a three-story building that was under construction at the
time of the earthquake. A plan of the second-floor framing is shown in Figure 3-7. The column and
beam orientation shown in Figure 3-7 suggests that the framing system is much stiffer and stronger in
the direction perpendicular to the street (parallel to the y-axis of Figure 3-7), assuming that similar
rebar are used in all beams and columns in this building. The ratio of total column area to plan
footprint in this building is 1.3%.
a. north-south elevation
Common slab systems include the one-way and asmolen slabs. One-way slabs range in
thickness from 80 mm to 120 mm, and span distances up to 4 m. For longer spans and heavy loadings,
the one-way, or asmolen, joist system is typically used. This system is composed of one-way joists that
are formed by hollow clay tile blocks; the slab between the joists is cast directly atop the blocks. The
thickness of the asmolen slab is usually 300 mm (200 mm block and 100 mm slab) or 320 mm (250
mm block and 70 mm slab). Figure 3-11 shows the underside of three asmolen floor systems.
Moment-resisting frame construction fared poorly during the Izmit earthquake. According to
official estimates, more than 20,000 moment-frame buildings collapsed, and many more suffered
moderate to severe damage. Three- to seven-story apartment buildings were hard hit, although many
had been constructed in the past 20 years. Many of the collapses are attributed to the formation of soft
first stories that formed as a result of differences in framing and infill wall geometry between the first
and second stories, the use of nonductile details, and poor quality construction.
Figure 3-12 shows the collapse of six moment-resisting frame buildings in a village on the
outskirts of Gölcük. Every moment-resisting frame building on this street collapsed, and 122 people
died in these buildings. Nonductile details were observed in every (collapsed) moment-frame building
on this street.
Figure 3-12 Collapse of moment frame buildings, Gölcük Figure 3-13 Collapse of moment-frame and wall
buildings, Adapazari
Similar collapses were common in the epicentral region. Figure 3-13 shows the extent of the
destruction in Adapazari, a major city approximately 10 km from the line of rupture. Many of these
buildings were constructed with hollow clay tile infill in the frames perpendicular to the sidewalk.
Because the buildings often housed shops and commercial space in the first-story, glass panels and not
hollow clay tile infill walls were placed between the first story columns adjacent to the sidewalk, but
tile infill was used in the upper stories. Such an arrangement of tile infill created stiffness
discontinuities in these buildings, which may have contributed to their collapse.
Figure 3-15 Collapsed and damaged moment-frame buildings on the Gölcük navy base
a. Collapsed building b. View of failed first story
“Near-field” is the common term used to define the zone within 5 to 10 km of a major fault.
Earthquake shaking in the near field is often severe, as illustrated by recorded ground motions obtained
from the Northridge (1994) and Kobe (1995) earthquakes. Widespread collapse of older construction is
to be expected in the near field due to the intensity of the earthquake shaking. Although many moment-
frame buildings in the near field of the Izmit earthquake suffered gross damage or collapsed, some
fared surprisingly well. Shown in Figure 3-17, the four-story moment-frame building is sited within 2
m of the line of rupture and yet suffered no visible damage despite 1.2 m of horizontal offset and 2.35
m of vertical offset on the fault. Figure 1-2 shows another view of the same building. The reasons for
such good performance of this 10- to 30-year-old frame building with masonry infill are unknown.
b. thirteen-story building
Figure 3-19 Varying degrees of damage to infill masonry walls
Damage to infill masonry walls was concentrated in the lower stories of buildings because of
higher demands on the strength of the moment-frame-infill wall system. Figure 3-20 illustrates the
distribution of damage to infill walls in two buildings, one near Gölcük, and one in Degirmendere. In
these buildings, the lateral stiffness of the masonry infill walls is likely of the same order or greater
than that of the moment-frames. For these buildings not to collapse following the failure of the infill
walls the moment-frames must have possessed significant strength and some limited ductility.
Figure 3-21 shows two views of a collapsed apartment building in Gölcük. The first two stories
of this building failed completely but damage in the upper three stories was limited. The long infill
walls in the upper three stories have significant elastic strength and stiffness—probably much greater
stiffness and strength than the moment-resisting frame. If the infill walls in the upper three stories of
the building are indicative of the infill in the failed stories, the first- and second-story infill walls likely
played an important role in the collapse of the building. The brittle fracture of the first- and second-
story infill masonry walls would have overloaded the nonductile first- and second-story frame
columns, resulting in a complete failure.
b. view of infill wall perpendicular to sidewalk
The first two stories of the building in Figure 3-22 collapsed. The infill masonry walls and
moment-frame construction in the third and fourth stories (first and second stories of the collapsed
building) suffered major damage. Damage in this building reduced with increased height above the
sidewalk. Failure of the infill masonry in the first and second stories of the building likely precipitated
the collapse of the building.
Irregular placement of infill masonry walls can produce discontinuities of stiffness in moment-
frame buildings. Consider the building in which the moment-frame is both flexible and weak by
comparison with the upper stories (Figure 3-23). In the first story of this building, infill masonry walls
are present in the back face of the building and in the two faces perpendicular to the sidewalk. The
front of the building was open in the first story. The lateral stiffness of the building was likely large in
the direction perpendicular to the sidewalk and much smaller parallel to the sidewalk. Deformation is
concentrated in the first story of this building, parallel to the sidewalk, due to the weakness and
flexibility of the moment-frame and the lack of infill masonry in the front of the building. The first-
story columns in this building were badly damaged and likely close to failure.
Figure 3-22 Failure of two stories of moment-frame building with infill masonry
3.5.1 Beams
The reconnaissance team observed little damage to interior moment-frame beams because
columns were generally weaker than beams. One type of beam damage is shown in Figure 3-24. The
building in this figure suffered a partial story collapse because the fault ruptured beneath the building.
The beams shown in the frame elevation were forced to accommodate the partial collapse and were
badly damaged at the beam-column connection due to slip of the smooth longitudinal beam rebar. In
many cases, beam bottom rebar was inadequately anchored in and through beam-column joints.
Figure 3-24 Damage to a nonductile reinforced Figure 3-25 Failure of lap splices in a moment frame
concrete beam conncetion
3.5.2 Columns
The majority of moment-frame component failures were in columns and were due to (a) the use
of nonductile details and unconfined lap splices, (b) excessive beam strength, and (c) interaction
between the columns and the infill masonry walls.
Lap splices in moment-frame columns were typically made immediately above the floor
framing or the foundation. The photograph of the exposed lap splice of Figure 3-25 is from a moment-
frame building in Adapazari. The lap splices in this column were approximately 35 bar diameters in
length and were located in a plastic hinge zone. Widely spaced transverse ties with 90° hooks were
used in this column; no cross ties were present. The 90° hooks on the ties opened during the
earthquake, and the limited strength and confinement afforded by the ties were lost.
Shear reinforcement was lacking in most damaged columns observed by the reconnaissance
teams. The transverse tie details of Figures 3-26, 3-27, and 3-28 were common, namely smooth rebar,
widely and often unequally spaced ties (200 to 250 mm), and 90° hooks. The wide spacing of the ties
resulted in shear failures (Figure 3-26), buckling of longitudinal rebar (Figure 3-27), and poor
confinement of the core concrete (Figure 3-28).
Figure 3-27 Shear failure of a moment-frame blade Figure 3-28 Lack of transverse reinforcement in moment-
column frame column
Shear failures in short columns were common. A typical example of such a failure is a school
building in Adapazari, shown in Figure 3-29. The damaged column of Figure 3-29b is shown in Figure
3-29a in a blue circle.
Column-infill masonry wall interaction resulted in severe damage to and failure of many
moment-frame columns. Consider the building east of Gölcük that is shown in part in Figure 3-30. The
infill hollow clay tile masonry on each side of the central column in this figure failed completely and
the column hinged at each end. The column to the left of the central column was captured
approximately 1 m above the floor by the residual hollow clay tile. The shear cracks that were
observed in this captive column formed in the column at the top face of the remaining infill masonry.
Figure 3-31 Concentrated damage at ends of moment-frame columns due to excessive drift
Figure 3-36 is a photograph taken in a building under construction in Adapazari at the time of
the earthquake. Severe damage in the beam-column joints is evident, but horizontal transverse ties in
the joints maintained the integrity of the joints. (See Section 3.7.3 for a more complete description of
this building.)
Figure 3-33 Damage to moment-frame Figure 3-34 a. Damage to moment- Figure 3-34 b. Damage to moment-
columns frame beam-column joints frame beam-column joints,
reinforcement in joint
Figure 3-35 a. Building collapse due to failure of beam- Figure 3-35 b. Damage to one beam column joint
column joints
Figure 3-36 is a photograph taken in a building under construction in Adapazari at the time of
the earthquake. Severe damage in the beam-column joints is evident, but horizontal transverse ties in
the joints maintained the integrity of the joints. (See Section 3.7.3 for a more complete description of
this building.)
Damage to such systems was widespread. Figure 3-37 shows typical damage to the asmolen
floor system. Deformation of the joist-beam framing led to sections of the hollow clay tile formwork
dislodging and falling to the floor below. Although failure of the hollow clay tile blocks in the floor
system is not considered structural damage, falling tile blocks constitute a hazard to life.
The building shown in Figure 3-40 experienced damage at the stiffness discontinuity in the
shear wall. The fault ruptured directly beneath this building. The limited damage in this instance
constitutes excellent performance.
Figure 3-40 a. Shear wall building damaged due to fault Figure 3-40 b. Close up of damage to a shear wall
rupture
The reconnaissance team toured a number of buildings that would be classed as dual wall-
frame systems in the United States. However, because design provisions for such systems did not exist
in Turkey prior to 1997, these buildings would have been designed as either shear walls or moment-
resisting frames. The most significant damage observed by the team in a dual wall-frame building is
shown in Figure 3-41a. The wall and first-story exterior columns shown (Figure 3- 41b) failed and
shortened. These components displaced out of the plane of the wall, as seen in Figure 3-41b.
Figure 3-41 a. Collapse of dual wall-frame five story Figure 3-41 b. Close-up of failure of the shear wall and
building, Adapazari perimeter columns
Figure 3-42 a. Damaged wall-frame building due to Figure 3-42 b. shear wall settlement
ground failure and wall rotation
Another example of damage to beams and columns in a dual wall-frame building is shown in
Figure 3-42. No cracks were observed in the shear wall, but the right end of the wall settled
approximately 0.5 m (Figure 3-42b) due to bearing failure of the supporting soils. Although the shear
wall was likely sufficiently stiff to protect the nonductile frame, the rotation at the base of the shear
wall and settlement of the footings beneath the moment-frame columns contributed to the failure of the
first-story columns.
Modern standards for the seismic evaluation of buildings (FEMA 1997) dictate decisions
regarding system response using information on component response. For the system performance
level of collapse prevention, system failure is linked to the first failure of a component (typically
measured in terms of deformation demands or demand-capacity ratios). Such correlation of system and
component response is misleading and often overly conservative if the seismic and gravity load-
resisting systems are redundant. One objective of the reconnaissance team was to gather information
related to limiting states of response of building systems, with an emphasis on the limit state of
collapse prevention. The following sections describe in some detail the performance of four buildings:
A through D. The first three buildings (A, B, and C) sustained severe damage to critical components
but did not collapse. The fourth building (D) performed poorly but as a result of ground failure.
3.7.1 Building A
3.7.1.1 Description
Building A, shown in Figures 3-44 and 3-45, was
located at the eastern outskirts of Gölcük. Much of
the first story of this moment-frame building (not
seen in Figure 3-44) was located below grade. The
grade level sloped down from the front to the back
of the building. A sketch of the first-floor plan of
the building is shown in Figure 3-46. Most of the
hollow clay tile infill masonry failed during the
earthquake but some remained intact at the rear of
the building in the sixth story (see Figure 3-45).
3.7.1.2 Component Failures
Structural damage was concentrated in the first-story columns at the front of the building
Figure 3-47) and around the stairwell at the rear of the building (Figure 3-48). Nonductile detailing
was evident in each damaged component viewed by the reconnaissance team.
The staircases in the rear stairwell were cast integrally with the exterior columns. The landings
were located approximately 1 m below the beam-column joints (Figure 3-48). No transverse
reinforcement was present in these joints. The lateral support provided by the landings and the
staircases resulted in short column construction and led to shear failures immediately above the
landings. Figure 3-48 shows severe damage to the staircases that suggests that the staircases resisted
significant lateral forces during the earthquake via strut action. The lateral stiffness of the staircases is
evinced by the damage they suffered but likely was not included in the earthquake analysis of the
building (which is also common practice in the United States).
The distribution of damage to columns in the first story is shown in Figure 3-46. Figures 3-49,
3-50, and 3-51 show column failures. Nonductile detailing is evident, including widely spaced
perimeter transverse ties with 90° hooks and no cross ties, and lap splices located at the floor level with
no confining transverse reinforcement.
As shown in Figures 3-46 and 3-47, the first and third rows of columns were badly damaged
but the second row of columns suffered no significant damage. All columns in the first three rows were
the same size; rebar in the first and third rows of columns were essentially identical. If the interior
columns in the first row failed initially, conventional approaches would suggest that lateral forces were
redistributed to other stiff components (including the second row of columns) and gravity loads were
transferred to the undamaged columns in the first and second rows. The increase in the gravity and
earthquake effects should have been greater on the second-row columns than on the third-row
columns, yet the columns in the third row failed and the columns in the second row were undamaged.
New knowledge regarding the transfer of lateral loads and gravity from failed components to other
components of a building frame is needed to obtain accurate estimates of building performance.
Although several columns in the first story of the building failed in shear and axial
compression, the building did not collapse. Clearly system response cannot be judged on the basis of
the most highly loaded (forces or deformations) component in the building, as is the practice in FEMA
273, NEHRP Guidelines for the Seismic Rehabilitation of Buildings (FEMA 1997). The gravity load-
resisting system of the building did not collapse for a number of reasons that include (a) frame action
in the stories above the damaged columns and (b) residual axial-load capacity in the heavily damaged
columns.
After the columns in the first row failed in shear and shortened, the slab and beam framing
deflected in the shape of a catenary (see the sag in the floor slabs in Figure 3-44) and gravity loads
were carried to the adjacent undamaged columns by axial tension in the beams and slabs. Vierendeel
truss action in the upper stories also likely transferred gravity loads to adjacent undamaged columns.
Provision for such redundancy in framing systems would reduce the likelihood of building collapse
and substantially uncouple system-level response from component-level response. The catenary and
Vierendeel truss mechanisms may be very effective in stabilizing the structure when interior columns
are lost. To ensure that beams and slabs are able to maintain catenary deflections, bottom
reinforcement should be continuous through any columns that may fail under lateral loads.
Recent studies (Moehle et al. 2000) have shown that columns heavily damaged in shear are still
capable of supporting axial loads. Residual axial strength in these columns would reduce the need to
redistribute gravity loads as described in the previous paragraph. The failed columns in the first row
were squat so that after failure in shear, the upper segments of the columns bore on the lower
segments, albeit not concentrically. (Contrast this behavior with that described earlier for narrow
columns; see Figure 3-32). The core concrete in the failed columns in the third row continued to carry
gravity loads after the earthquake because the cores of the columns remained partially intact. The use
of transverse reinforcement in the amount needed to keep the core of a column intact at large
deformation would further reduce the likelihood of building collapse.
3.7.2 Building B
The loss of one or more components in a moment-frame building can substantially modify the
magnitude and behavior of the remaining components. Assuming that the location(s) of the failed
component(s) are known, nonlinear methods of analysis can be used to evaluate the forces and
deformations in the damaged building frame. Two challenges with such analysis are (1) identifying the
number and locations of components to be removed from the mathematical model and (2) including
the effects of column failure and load redistribution.
Procedures for selecting the number and locations of components to be removed from the
mathematical model have not been developed. The number and locations will vary as a function of the
earthquake histories used for analysis and evaluation. Demand-to-capacity ratios (deformations for
ductile actions and forces for nonductile actions) could perhaps be used to identify combinations of
components for removal from the mathematical model. Two approaches could be used to assess
system response following the failure of selected components: (1) remove the components from the
mathematical model before analysis and (2) remove the components from the model during the
analysis when deformations or forces, or demand-capacity ratios exceed a threshold value. Approach 1
is more conservative than Approach 2. Approach 1 could be used with nonlinear static or dynamic
analysis. Approach 2 would be used only with nonlinear dynamic analysis.
The rapid loss of a column or beam can lead to a dynamic amplification of the gravity loads
that are transferred to adjacent components. Procedures for calculating the amplification factor are not
available at this time. Studies are very recently completed at PEER by Rodgers and Mahin on steel
moment-frame buildings and by Elwood and Moehle on nonductile reinforced concrete moment-frame
buildings to evaluate the effect of component failure on system response.
3.7.3 Building C
One of the failed columns at the rear of the building had a very steep shear crack (Figure 3-56)
that suggested that the column was carrying high axial loads. Following a more detailed inspection of
the building, the reconnaissance team concluded that the column was part of a two-story addition,
which was separated from the five-story building by an expansion joint. No plausible explanation for
the steep shear crack is proposed. The beam-column joints at the top of the first-story columns on the
eastern façade were damaged but did not fail because transverse reinforcement was provided in the
joint region; see Figure 3-57. The beam-column joint at the north-east corner of the first story was
heavily damaged, as shown in Figure 3-58, but continued to carry gravity loads.
Figure 3-55 View of retail space in the first story of the
Building C
To predict incipient collapse of a building or to evaluate buildings for the performance level of
collapse prevention, new information on how gravity loads are supported in buildings with severely
damaged or failed components is needed. The severely damaged interior columns of Figure 3-55 could
support little or no gravity load. This observation suggests that much of the gravity load in the building
must have been distributed to the perimeter first-story columns by Vierendeel truss action in the upper
stories. Many of these perimeter columns suffered damage to their beam-column joints, but the use of
transverse reinforcement prevented joint failure and gravity load resistance was maintained.
Although the residual drift of the first story of the building adjacent to the front sidewalk was
approximately 5%, the P - ∆ effects did not lead to collapse of the building. Three factors probably
contributed to the stability of the building. First, the shear wall near the stairwell (see Figure 3-55),
although heavily damaged, likely had significant residual lateral stiffness and strength. Second, the
axial loads in the columns were low as a percentage of fc'Ag. Third, the residual drift at the rear of the
five-story building was much less than 5% and the framing at the rear of the building may have
partially stabilized the building.
3.7.4 Building D
3.7.4.1 Building Description
Building D was a six-story moment-frame building located in the center of Adapazari. An
elevation of the building is shown in Figure 3-59. Based on similar construction of the same age in
Adapazari, the foundation for Building D was probably a mat or raft with a thickness of approximately
1 m.
At the time of this writing, building (system) response is often judged on the basis of the most
highly damaged component in the building. Clearly, this approach, although conservative, is neither
accurate nor cost effective. Poor behavior of one or two random components does not necessarily lead
to poor system behavior, although poor behavior of one or two key components may lead to system
collapse if mechanisms for redistribution of gravity loads do not exist in a building.
Much remains to be learned about the collapse of buildings and the design of buildings on soils
prone to liquefaction or failure. Research on the following topics is needed to improve analysis,
evaluation, and design procedures to ensure with high confidence and low cost that buildings will not
collapse.
1. Triggers for axial load failure of ductile and nonductile reinforced concrete columns under
combined loadings based on large- or full-scale test data.
2. Mechanisms for redistribution of gravity loads in the event of component(s) failure, and
characterization of gravity-load amplification effects due to component failure.
3. Analytical tools for predicting component strength and stiffness loss under combined loadings based
on evaluation of large-scale experimental data.
4. Procedures for eliminating components from mathematical models to simulate component failures.
5. Large-scale 3-D earthquake simulator testing of buildings with weak and brittle components to
validate the analysis, evaluation, and design procedures developed in 1 through 4 above.
Editors note: This text follows as closely as possible from the paper copy of Chapter 3 “Reinforced Concrete Frames and
Wall Buildings” by H.Sezen et al. Structural Engineering Reconnaissance of the Kocaeli (Izmit) Turkey Earthquake of
August 17 1999. Berkeley: Pacific Earthquake Engineering Research Center, (PEER Report 2000-09), December 2000. A
few figures have been substituted where color originals did not exist. Minor text and layout modifications were made.
Higher resolution images of all images are available through EQIIS image database http://nisee.berkeley.edu/eqiis.html by
searching under the “Izmit (Kocaeli), Turkey earthquake, Aug. 17, 1999”
C. James, 2001